Hi, Biostars:
I'm recently doing transcription factor (TF) footprinting using DNase-seq data, the basic work procedure is from Prof. John A. Stamatoyannopoulos's lab.
The initiation step for footprinting is to correct the DNase cleavage bias, which has been extensively researched by his lab (Lazarovici et al., PNAS, 2013) and other labs (Koohy et al., Plos One, 2013; He et al., Nat Med, 2014; Baek et al., Cell Reports, 2017). As far as I know, packages often correct DNase bias using the 6mer sequence frequency, because the DNase motif is 6bp.
However, Lazarovici et al. reported that DNA methylation can significantly affect the DNase cleavage rate, I'm curious that why these packages did not take this parameter into consideration when correcting bias? Actually, the recent large-scale TF footprinting work from John's lab (Vierstra et al., Nature, 2020) also only corrects the sequence bias, without the DNA methylation effect.
Whether this is due to that the correction of DNA methylation needs the true DNA methylation information on corresponding cell type, which is not practical? As Lazarovici mentioned in his mannual, he calculated the 6mer bias from naked DNA, and use this information for correcting the chromatin dataset. Will it be possible to calculate DNA methylation-considered 6mer bias from the naked DNA using well-known cell types as a reference, then correct other cell types?
Thanks for your time, I appreciate any comments and suggestions on this question.
